Home > Databases > Open Source: Good for the OS, But Not for the DatabaseOpen Source: Good for the OS, But Not for the Database Eugenia Loli 2004-10-09 Databases 30 CommentsLarry Ellison embraces Linux, but don’t expect open source databases from Oracle any time soon. (Part Two of six part interview). About The Author Eugenia LoliEx-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.Follow me on Twitter @EugeniaLoli 30 Comments 2004-10-09 7:15 pm Releasing the Oracle server source to the community would likely serve little purpose, since most people who are buying Oracle are also buying the expensive support packages as well. While I fervently believe in open source, sometimes the notion that you can fix problems by having access to the source is just comedy. No one with a timely and serious database problem is going to be digging through 100k++ lines of code. 2004-10-09 7:28 pm That is an usual thing for small customers, small to medium companys, but imagine a google like company, or half of it, and imagine they had to pick oracle, it would have costed …It is eazyer for them to have inhouse staff for doing patches and other things.But opensource gets also into the segment of small companys because it is cheap and cheap is more important then the feel they get from buing a closedsource one.And opensource does a good job, it is a perfect sollution for any kind of client that is interested. 2004-10-09 8:24 pm done the asbestos, i see a flamewar comeing why bother open sourceing oracle anyways? there is mysql and postgresql atleast otut there that is open source… 2004-10-09 8:40 pm no one thought an open souce operating system would be able to compete with the big boys back in 1991. But a little over 10 years later and it’s 2004 and linux knocking around on the low end of the market (high end if you consider google). Every cake takes time to bake which basically means give it time. AFAIK, the open source databases are not used for the transactional systems at banks or airlines. But I am quite certain that by 2012 there will at least one or two small banks ($300 million in assets) that will be using open source databases for their transactional systems. It probably won’t be until 2018 that you see one of the big boys start (Citigroup, BofA, Wells Fargo, etc.) to catch on.Don’t expect a bank to “bet” it’s business on an open source database just yet, but it will happen. Again just give it some time.BTW: you pay far far more in support and people costs (ie a good DBA and sysadmin) than you usually do in database costs. And then there are application developers that the bank hires to write custom apps. 2004-10-09 8:40 pm Apologies for stating the obvious, but MySQL and PostgreSQL aren’t exactly up to the calibre of Oracle for large scale enterprise deployment. Just the complexity of the SQL dialect is a great enough advantage alone. 2004-10-09 9:09 pm Ellison made some comments about mySQL not handling transactions, but PostgreSQL handles transactions, it has referential integrity, on-delete, etc. He never says why someone should use Oracle over PostgreSQL. Frankly, mySQL is getting these features slowly but steadily. Of course, most people refuse to use them since using them entails using InnoDB tables and therefore a performance hit.Even more, PostgreSQL has shadowed Oracle and it is probably much greater competition than mySQL. I think it would have been a lot harder for him to discount an OSS database that aimed for the same market as Oracle like PostgreSQL does. 2004-10-09 9:40 pm If OSS databases continue to grow at the rates they are currently growing at, then I can see Oracle changing its tune…but don’t expect to see it happen overnight; give it some years and it may happen….Michaelhttp://www.phantasyrpg.com/main.php?view=9898 2004-10-09 10:57 pm Ellison on MySQL:”It has pretty good query capabilities, but it’s not very strong transactionally. I would say that they’re not as strong from a securities standpoint or a liability standpoint or a scalability standpoint.”I think the liability issue is greater than the technical issues. Many companies go with Oracle or DB2 etc. out of fear. The IT manager doesn’t want to lose his data/job because his database barfed. But that doesn’t mean that the companies with staff who know their stuff aren’t going to find that free databases can more than meet their needs.My own business is utilizing a mix of free software and proprietary systems. If I had been running this business 4 years ago, my licensing costs for software would have been in the region of $100,000 per server more than they are now, and most of that benefit has come from free software or the price-cutting that the proprietary companies have done because of free software. If it wasn’t for free software I don’t think I would have taken the plunge, and I’d still be another in-house developer.I’ve worked for companies that spend as much on their annual licensing and consulting fees for Oracle and Websphere as they have spent on the salaries of their entire internal development team (I’d rather invest in my own development team). These companies didn’t need Oracle or Websphere, but the IT managers would rather trust to a brand and a license. They like to say “we can sue Oracle if something goes wrong”, but they never would. Even after they got nothing from Software Assurance, they didn’t sue Microsoft.There is clearly a competitive advantage to using open source. Those People just don’t get it (yet). If people are prepared to do some research and testing, they can find out for themselves what the limitations of free software are, and see if they are suitable for them. These businesses with the competitive edge will supplant those without it; eventually, Oracle’s sales will suffer.One organisation has a product to enable people to transition off Oracle to Firebird:http://www.janus-software.com/fb_fyracle.htmlAccording to their web-site, Firebird is the only open source db that is up to this:”You will find that only one open-source database comes close to matching the leading enterprise systems on features: Firebird.” I don’t necessarily agree… PostgreSQL also has some great features (some of these are simple and I have been able to write for myself using things like triggers, but I wish Firebird would adopt some of those features e.g. metadata changes under concurrent versioning).Sure some companies are going to need the clustering/replication features of a database like Oracle. Firebird doesn’t offer the kind of replication I need for my scalability design – as that was the missing factor I built it for myself. Now I can add new servers and the software cost of each additional server is less than $10,000 – and I’m using best of breed for everything. My expansion costs are about 10% of what they would be in Larry Ellison’s vision. 2004-10-09 11:46 pm “But opensource gets also into the segment of small companys because it is cheap and cheap is more important then the feel they get from buing a closed source one.” Screw “cheap” companies want stuff that works. The cost doesn’t matter a damn bit if the stuff doesn’t work. 2004-10-09 11:54 pm RE: ” Screw “cheap” companies want stuff that works. The cost doesn’t matter a damn bit if the stuff doesn’t work. ”on the other hand companies don’t want to be overburdened with highly expensive “sacred cows” of an application and/or OS either, hence to move to Linux & OSS… 2004-10-10 12:08 am See subject. I’m just saying we shouldn’t be surprised if a CEO thinks he’s product is going to compete well. Larry would be the last person to admit otherwise, at least to the public.I think he was right to point out the importance various corporations have to Linux. Open Source can be used as a vehicle for otherwise competing companies to work together for their mutal good. Eclipse is a clear example of that. 2004-10-10 1:53 am You can get an excellent commercial grade, enterprise class database for Linux free from Sybase:http://www.sybase.com/linuxpromoThe limits are:For example, users can only use the free download version on systems that have one CPU, no more than 5GB of data storage, and a maximum 2GB of memory.Other than those limits it is full featured and you can use it for any purpose. They have really nice Java and XML integration. 2004-10-10 5:14 am I see these stories and then comments like this and am baffled, wondering to myself..”why isn’t anyone talking about Firebird”?Firebird is a fully transactional database with an *extremely* light footprint, is extremely fast, scalable, supports sprocs, triggers, and literally any enterprise-level feature you would want or need…has a history of nearly 20 years in the market (formerly Borland InterBase)…!!!!Firebird, I use it, I love it, and I’ve never paid a single penny for it…http://www.firebirdsql.org 2004-10-10 5:48 am OSS/free software is good for _____ but will _never_ be good enough for our specialized field of _____. Our product ____ is/will always be far superior than the OSS/free project _____.People have been saying this since OSS and free software have been around and eventually get proven wrong time and time again. These statements are almost always formulaic and have no sense in being reported on. Perhaps more news worthy is the fact he is saying anything about it which is a good clue that OSS/free software is already messing with his bottomline and at least some of his target market. 2004-10-10 7:24 am Has anyone read a M$ EULA?It pretty much states that they are not responsible for anything, and I bet DB2’s and Oracles are similar.If you lost critical data because of a bug in Oracle or DB2 I think you would be SOL, and that’s why backups are critcal.You could of course try to sue them, but if you agreed to the EULA it would be pretty tough to get them to pay.Look at all the money M$ wasted for companies by having all those security holes in it’s OS and letting in nasty viruses.It’s up to the company to ensure everything works before going into production with a DB. 2004-10-10 9:03 am That is so correct about FireBird.The main issue with it is of popularity, you can see dozens of LAMP,WAMP tutorials, but people disregard FB because of the fact that they are not aware of what a RDBMS is and what it does.They usually do INSERT … SELECT .. DELETE … UPDATE…., i mean even CREATE TABLE for them is too complex.They learn everything on the fly, they dont even know what SQL as a standard is.MySQL inmho could be replaced overnight by FireBird in most of the php applications that we see(few of them really use mysql specific stuff).But no on knows about Trasactions,StoredProcedures,Triggers,Views … if they have their SELECT for their fetch_stuff php function then there is all they need.Thats why abstract systems like AdoDB for php for ex is so good.But who could make developers thing this why, if they ARE ON THE FLY? 2004-10-10 11:14 am Basically I’m agree. It’s only a matter of time. Relational database products are basically a “dead-end”. The market and the technology is mature, and the software are now being commoditized. I’m sure even Larry knows that. Oracle has been trying to branch off to enterprise applications more and more. 2004-10-10 11:24 am MySQL inmho could be replaced overnight by FireBird in most of the php applications that we see(few of them really use mysql specific stuff).Not that fast, my dear. 🙂 MySQL does have several features not existing in Firebird or other DBs.1) Full text indexing/searching (builtin one, which is very convenient).2) Then there’s spatial data types in 4.1 (I don’t recall IB/FB ever having those).3) GROUP BY … WITH ROLLUP.4) some functions like MD5(), SHA1(), GROUP_CONCAT(). I know many of those are only a matter of writing or finding UDF libraries, but stil…5) and when are they going to ship FB that supports longer index lengths? I’m kind of tired of waiting…And to top it all, Firebird is not a replacement for MySQL in all areas. In multiuser mode it’s harder to debug problems & monitor since it tends to be a black box.It certainly takes more than a day to convert some MySQL-based applications to Firebird (or other DB). 2004-10-10 11:26 am Forgot to add… Some of the abovementioned features do get used a lot. Full-text indexing and the various MySQL functions, for example. Many CMS or forum applications basically depend on MySQL’s full-text indexing to provide their searching capability. 2004-10-10 1:07 pm I dont try to compare features vs features, i was talking why a product that seems more mature is not penetrating the programmers space, then main issues that come to be i proposed to be:popularity.lazyness(as in interes to know the facts).lack of experience and knowledge. 2004-10-10 1:48 pm Because of bad mentality of Oracle, Sun, IBM and other big companies, Microsoft made money selling their cheaper (and worser) software. Oracle is very expensive and, except in a few situations, it can be replaced by Firebird or Postgresql without problems.Fortunately, linux and many free software projects are changing market. Death to stupid corporate mentality ! 2004-10-10 2:22 pm Well, it certainly seems that we’re facing a very interesting development for the next few years. In my opinion there are three promising and upcoming open source DBMS’es, and all with quite different approaches:– MySQL, which was intentionally designed to be the fast and extremely lightweight DBMS, is gaining on more and more features, but still lacks what I would consider extremely basic features: Views, Triggers, Stored Procedures, Check Constraints(!), more sensible schemas, etc. But it’s fast and has some nice features, including the text indexing that you talk about. However, it remains to be seen if it’ll stay as lightweight when all the new features are incorporated. You already have the InnoDB which slows things a bit down.– PostgreSQL, which started in the other end: Aiming at fulfulling almost every standard written. Even though it’s very compliant, it still lacks some features like Assertions (does any DBMS really have this?). It used to be very difficult to approach, especially because it was only (easily) available on *nix plattforms. That’s the reason why the website PHP teenagers didn’t experiment with it. This is changing with the upcoming version 8.0, and speed is also improving. And to help everything, the web-based admin system for PHP, phpPgAdmin, is in my opinion already better than it’s MySQL counterpart, phpMyAdmin. This will further help on PostgreSQL’s popularity for web developers – which is an important marked because so much really *starts* here.– Firebird, oh yes. A very exciting competitor among the three, and by far the most mysterious one to the masses. It seems that Firebird’s major drawback is not in the product itself. It’s the least “buzzed” DBMS of the three, and people in general know very little about it. I haven’t developed anything for it yet, and to be honest it’s not easy to find all the stuff that curious souls like me need in order to get started and give it a try.So, my point?My point is that these three DBMS’es are more competing with *each other*, than anyone is with Oracle. Oracle will still remain to be no. 1 for big-scale enterprises.The three open source DBMS’es are unlikely to overthrow Oracle at any time, but they are starting to get more and more similar in features and performance. This again will lead to a more unified stronghold against Oracle and DB2, and as this correlation increases, hopefully we’ll see that some old school and conservative people will raise their eyebrows.Or find themselves beaten by the next generation. 🙂 2004-10-10 3:08 pm Yep, fill in the blanks – and consider the source.If you’re sunw, then opensource is great for desktops (where it competes with msft), but opensource won’t work for servers (where it competes with sunw). If you’re oracle, then opensource is good all sorts of things – but not databases. If you’re msft then opensource isn’t good for anything. 2004-10-10 5:48 pm I think Oracle are safe, for the timebeing. Many organisations have an almost pathological devotion to Oracle, because it’s what they know. Many will use Oracle on Windows (yep, that’s right), even though the performance is crap and SQL Server runs much better on Windows (surprise) and Oracle runs much better on anything else. I’ve seen one company move expensively to Oracle 9i, in the middle of a company-wide totally critical project, and that was simply because Oracle were removing support for all previous versions!In the long-term, I can see companies like MySQL growing steadily and people providing support around PostgreSQL. Oracle is the next Sun, and it will actually be even worse for them because expensive databases are all they really do. They’re safe for now, but they need to have another good look at their business models and anticipate what will happen in the long-term. 2004-10-10 6:12 pm > I’ve seen one company move expensively to Oracle 9i, in the middle of> a company-wide totally critical project, and that was simply because> Oracle were removing support for all previous versions!Oracle 8.1.7 (prior than 9i) is supported until december 2007 2004-10-10 7:33 pm Oracle 8.1.7 (prior than 9i) is supported until december 2007Well, that was the information they were given or the impression they had (laughing out loud here). Unfortunately, some organisations are so set in their ways that you simply cannot tell them anything. 2004-10-10 7:41 pm Oracle knowns that commercial databases market is doomed.But applications are even more important than the databases, and that’s why they want to buy peoplesoft. If free databases are very performant but without any applications, they are useless. Applications are the only way to save databases sales. Currently, when you say “mysql”, people think “web, php”, not “crm”. 2004-10-10 7:48 pm Ellison made some comments about mySQL not handling transactions, but PostgreSQL handles transactions, it has referential integrity, on-delete, etc. He never says why someone should use Oracle over PostgreSQL. Frankly, mySQL is getting these features slowly but steadily. Of course, most people refuse to use them since using them entails using InnoDB tables and therefore a performance hit.Even more, PostgreSQL has shadowed Oracle and it is probably much greater competition than mySQL. I think it would have been a lot harder for him to discount an OSS database that aimed for the same market as Oracle like PostgreSQL does.heh… even if it can’t compete with Oracle today (but sincerely, who can? Microsoft SQL can’t, let alone MySQL), PostgreSQL is indeed the OSS DB that approaches Oracle’s caracteristics more closely.The more development is made around PostgreSQL, the more the following equation becomes realistic:(Linux/FreeBSD) : Microsoft = PostgreSQL : Oracle… we’ll see. 🙂 2004-10-10 9:39 pm and this fact scares the hell out of me. if there was lifes at stake no goverment in its right mind would allow for ms or any other company to put stuff in their EULA saying that they were not responsible when it comes to loss based on flaws in said product. thats like haveing a toaster blow up under normal use, wounding or killing the user and then the maker saying that the user was at fault as he didnt take the toaster appart to look for flaws (excuse me but i cant take that appart, your screws use a proprietary head!).ok we then get stuff like what is “normal” use. this is what have lead to labels like “surface hot while in use” on some household items. problem with that is that it have become leagaly allowed to be willfully ignorant of common sense.its a two way street. if no flaw in the software used lead to said lass of data then it should be written of as user error. but if its a reprodicible flaw that leads to said loss of data then the fix should be applyed to any of makers products where it shows up, no matter if its out of service rotation. either that or they open source the product the second they take it out of their support system. right now its a choice at gunpoint, update or have the potential bomb in a nice wrapping blow up in your face.this is why open source is so good, the ability for infinite support, no forced (or semi-forced) updates and a continual evolution of the product rather then a repeated revolution. basicly it becomes more like a living ecosystem where the software continualy mutate rather then getting a injection of genetic code every now and then to hopefully help the immunesystem.in fact, applying a eco-system like idea to modern computing may not be so far of. viruses and its derivatives are in fact correctly labeld. even more so now with the net as a spread vector. just like a patient with a air or contact vector illness is best of being isolated, a computer with a network transmistted virus should be isolated from said network if there are other computers that have the risk of getting infected. the problem is that the speed of infection is so high the a human cant do the detection and isolation.this is a problem that in fact impact any os out there. no mac, linux, bsd, windows, or any other os is immune. they are just diffrent kinds of os and therefor immune to problem of the others.we all have a degree of responsibility in this, users, sysadmins, developers, all. hunting for someone to blame just make us feel better, it solves nothing. even if you have a piece of digital text or a signed piece of paper that say that one side agreed that the other part in the agreement have no responsibility they still do, atleast moraly…some times i think that a drivers licence for computers s a good idea. but the people presenting it make it sound like a solution to solve all the problems. it just makes more people aware of the problems and responsibilitys. and as people have a bad habbit of showing, they ignore those responsibilitys when they feel like it (drunk driveing, failing to install the latest patch and so on). it will maybe make the sysadmins that suggested it sleep better at night…maybe the best idea about be a computer that cant be directly infected. basicly the os and binarys are not stored in a RW enviroment. this removes the primary way for a virus to get a hold as all it takes to remove them is to do a reboot. this tho would require a whole freaking lot of ram in said box so that you could load in something of a cd or dvd and then remove it to load something else while the first application stays in memory. or maybe use a hd to temporarily store said application on. upon reboot (rather then sleep) the binarys on the hd needs to be verifiyed of said cd/dvd before the application is allowed to be used again…maybe what im describing is what ngscb is supposed to provide in combo with the tcpa (was that the right acronym?) chip. the ability to sign binarys… 2004-10-16 11:43 am Hmm, all interesting stuff… but in the end, this is about business!In my experience, CIO’s cover there royal behinds… its not about the best technology, the fastest, the cheapest, the best supported… these are questions that are asked naturaly, but in the end, it boils down to the CIO keeping his job IF things go wrong. And a CIO backing open source software is a sure way to a short career!I have worked with all the main stream commercial and free RDBMS’s, Oracle (my pet hate), MSSQL, Firebird (the most interesting), SapDB (Now MaxDB) MySQL and the rest. They all do what an RDBMS is supposed to do! They all have nifty bits that do particular things, and they all have strenghts and weekneses.Many of my clients, large and small, have a budget to spend… and they WILL spend that budget! Then, your into a marketing/selling game… and that is why M$ and Oracle have such a large presense in the market.It is also important to note that in a Boardroom servey it was discovered that of the 8 core functions within a boardroom that were ranked in order of importance by the CEO, CIO was seventh out of 8! Just above HR which was last. CIO’s and technology are not seen as key drivers of competitive advatntage anymore… the days of the CIO/Technology as king are over.It is interesting though, that several of my larger clients (including a 65000 seat user base in 120 countries) have now started open source initiatives. Note, these initiatives were not strarted because of TCO! They were started because big business in now begining to recognise that some of the best software being writen is comming out of the Open Source arena.